The paleo/real-food sphere had a mini freakout a few weeks ago when headlines started popping up that coconut oil is bad for you (for example: here). These headlines were precipitated by (trying not to use words like ‘the result of’ or ’caused by’ given what’s coming below) an American Heart Association (AHA) presidential advisory that was published in Circulation (an AHA publication). The article reviewed the literature on effects of reducing saturated fat and replacing it with unsaturated fat on cardiovascular disease. It had one short section on coconut oil, but coconut oil specifically was not a main point of consideration in the literature reviewed. The article ultimately concluded that reducing saturated fat and replacing it with polyunsaturated and monounsaturated fats reduces cardiovascular disease (heart attacks, strokes). Since coconut oil is high in saturated fat (and the specific studies related to coconut oil the article cited indicated that coconut oil increases LDL cholesterol, which the article cites as being a cause of cardiovascular disease), the AHA advises against using coconut oil.

Polyunsaturated and monounsaturated fats, which the AHA recommends, include canola oil, corn oil, soybean oil, peanut oil, safflower oil, olive oil, and walnut oil, among others.


My initial reaction, like that of many others who eat the way I do (or the way I try to), was to sort of roll my eyes and wonder how they’re manipulating or selectively interpreting the evidence this time. But. Right there with my initial reaction was the thought, “But what does the article actually say? What evidence does it cite? Are there methodological flaws to that evidence, or is it pretty strong?” The truth is that I don’t really know. I assume there must be something wrong with it, because I feel pretty confident in these two things:

  1. Naturally occurring, real food is good for you. This includes naturally occurring, saturated fat, like that found in coconut oil, lard, and butter, which humans have had access to and subsisted on forever and ever, amen.
  2. Industrially-produced ‘vegetable’ and seed oils, like canola oil and soybean oil, are bad for you.

And part of me feels like I don’t care what the evidence says, for the following reasons:

  • The two points above make sense. It makes sense that what humans ate up until 50-100  years ago and what occurs naturally in nature is good for you. And it makes sense, especially considering the rise in chronic disease over the past 50 years, that the industrial production of food generally is not good for you.
  • I am becoming increasingly skeptical of the body of scientific literature that exists, at least around food and nutrition. Corporate interests and funding are so intermingled with so much research. In this case, big agriculture, like corn and soybean growers.

That said, I am a generally rationale person who wants evidence. I have a masters of public health – not a science degree, but I’m trained in epidemiology and evidence-based thinking. I insist on others having data and evidence to back up their claims.

And, I’m married to a very critical-thinking evidence-based medicine-supporting physician who cares about knowing the evidence.

I’ve never been good or comfortable at making strong arguments off the cuff, especially without really knowing what I’m talking about. So – new personal research question:

  • What does the evidence say regarding health outcomes of eating different types of fats, from different sources? What are the strengths and weaknesses of the evidence?

So, I’m going to embark on a little personal journey to really try to understand the AHA’s article, the evidence it cites, the other evidence that’s out there that it doesn’t cite, and really try to wrap my arms around this. Some people (like the authors of AHA’s paper, presumably) conduct literature reviews as part of their full time jobs, and I’m not sure how much time and energy I’ll really be able to put into this. Because I know I’m going to want to go deep. But we’ll see. I’m going to start by reading the article closely, summarizing what it says, and determining what additional questions it raises for me that I might want to look into. Starting now.

Summary of Article:

Okay, as it turns out, I’ll need to do this in chunks. Below is a summary of the first 6 pages, with the additional questions I have (in italics). I’m not answering them now – just taking note of where I want to look into things more.


  • Purpose: Review and discuss the scientific evidence on the effects on CVD of dietary saturated fat and its replacement by other types of fats and carbohydrates
  • Rationale for decreasing saturated fat:
    • Saturated fat raises LDL cholesterol.
      • How well-established is this?
    • LDL cholesterol causes atherosclerosis (plaque buildup in arteries).
      • How well-established is this?
    • Populations with low saturated fat have low rates of CVD.
      • I can believe that this is well-established, but it doesn’t necessarily mean that populations with high saturated fat have high rates of CVD.
  • Current AHA guidelines:
    • Keep saturated fat to 5-6% of daily calories if you have elevated LDL and under 10% for the general population.
    • Replace saturated fats with poly and monounsaturated fats (I’ll use PUFA and MUFA from here on).
  • Summary of evidence: Randomized controlled trials and observational studies both have come to different conclusions regarding the effects of reducing saturated fat on CVD. However, RCTs and prospective observational studies in which saturated fats were replaced by PUFAs rather than carbohydrates (usually refined carbs) show reductions in the incidence of CVD whereas studies that replace saturated fat with carbohydrates do not.

Then the article discusses RCTs that looked at replacing saturated fat with PUFAs.

  • First, AHA looked at 3 meta-analyses looking at this question. It seems that within those 3 meta-analyses, 10 studies were identified. AHA chose four of those ten that it deemed to be the highest quality, having met the following criteria:
    • Compared high saturated fat with high PUFA intake
    • Did not include transf fat as a major component
    • At least 2 years of sustained intake of diets
    • Controlled dietary intake of the intervention and control groups
      • How did they do this? Is this reliable?
    • Proved adherence by objective biomarkers such as serum cholesterol or blood or tissue levels of PUFAs.
      • I’ll say right here that I am super confused about using serum cholesterol as a biomarker demonstrating adherence to a low saturated fat diet. It must be pretty well-established that replacing saturated fat with PUFA reduced serum cholesterol. Is it? Above, they said that high saturated fats increase LDL cholesterol. Is this the same thing? If saturated fats increase it then PUFAs decrease it? They use this biomarker a lot… sometimes seeming to use it as a result, sometimes as evidence that the control group actually adhered to a high PUFA diet. I’m just confused.
    • Collected and validated info on cardiovascular or coronary disease events
      •  So, a general question I have is how important are these qualities? If a study didn’t have them, is it right to not consider as high-quality evidence?
  • Study 1: Wadsworth Hospital and Veterans Administration Center in LA.
    • Date of published study: 1969
    • Double-blind.
      • Need to look up how this was done double-blind. They don’t explain.
    • N=846 men. 30% had CVD.
    • Diet:
      • The two groups were served meals in separate dining rooms at the center.
      • Control – usual diet (I assume, it doesn’t specify).
      • Experimental – Replaced saturated fat with PUFAs: corn, soybean, safflower, and cottonseed oils
      • Adherence measures\d by ‘enrichment with linoleic acid in blood, adipose tissue, and atherosclerosis specimens.’
        • I assume these are good biomarkers…. I don’t know.
    • Duration: 8 years
      • Were all the same people in it for 8 years? What was the average duration of individuals in the study?
    • Results:
      • Serum cholesterol reduced by 13%.
        • Is this a result or a biomarker that the experimental diet was adhered to?
      • Statistically significant 34% reduction in CVD, myocardial infarction (MI aka heart attack), sudden death, or stroke.
        • I believe this is measuring whether ANY of these outcomes happened. Looked at individually…:
      • 20% fewer MI or sudden death events, but not statistically significant
      • 41% fewer men with strokes (p=0.055)
        • This is on the edge of being ‘statistically significant’ if you’re going for a p-value of <5, but it’s close, so I think this is reasonable.
      • 31% reduction in CVD events, statistically significant.
        • So, this seems to be the main driver of the 34% reduction in any of these events above.
        • Also, how are ‘CVD events’ defined? What is a CVD event that is different from the other ones being measured?
  • Study 2: Oslo Diet Heart Study
    • Date of published study: 1970
    • N=412. Men that had had a myocardial infarction (heart attack)
    •  Diet
      • Control – their ‘usual high-saturated fat diet.’
        • Was this confirmed? Did they indeed have high saturated fat diets?
      • Experimental – participants and wives taught how to select and prepare foods low in saturated fat and high in PUFA vegetable oils.
        • Again, was this monitored? Was it confirmed that it was a) lower than the control group and/or b) lower than what they were eating before the study?
        • PUFA diet lowered serum cholesterol by 14%, confirming adherence
          • Here’s where I’m still confused. I think this is not a result. It’s meant to demonstrate that the low saturated fat diet was adhered to.
      • Duration: 5-year trial.
        • Again, were the same participants in it for all five years?
      • Results:
        • 29% reduction in recurrent MI and new cases of angina pectoris or sudden death (p=0.011)
          • MI reduced by 37% (statistically significant)
          • Angina pectoris reduced by 66% (statistically significant)
          • Sudden death the same in both groups
        • 25% reduction in MI or sudden death (p=0.05).
          • Are these just different ways of grouping all the potential outcomes?
        • 27% fewer cardiovascular deaths (p=0.09)
      • Study 3: British Medical Research Council
        • Date of published study: 1968
        • N=393, men that had had an MI.
        • Diet:
          • Control – ? Not specified in article.
            • Presumably their normal diet?
          • Experimental – participants told to drink soybean oil with fruit juice and also use it in cooking. Participants also counseled on reducing saturated fat.
            • Did they actually reduce saturated fat? Did they adhere to the protocol? Is a reduction of serum cholesterol evidence of this?? Clearly this is something I need to get figured out first.
          • Reduced serum cholesterol by 16%.
        • Results
          • 18% reduction in a recurrent coronary event (MI, angina, sudden death), but not statistically significant.
        • Study 4: Finnish Mental Health Study
          • Date of published study: 1972-1983 (three different publications)
          • N=1222 patients at 2 psychiatric hospitals
            • Also, two cohorts studied. One cohort had some participants that had evidence of coronary heart disease (arterial blockage starting) and some participants that did not. The other cohort only had participants that did not have evidence of coronary heart disease.
          • Diet:
            • In one hospital, high PUFA (mainly soybean oil) diet given first, followed by the sat fat diet. Each diet period lasted 6 years.
            • In other hospital, reverse.
              • But not all people were there the whole 6 or 12 years, right??
            • Serum cholesterol 14% lower in high PUFA diet groups.
            • Adherence also demonstrated by 3-fold enrichment of linoleic acid in adipose tissue.
          • Results
            • Coronary heart disease death reduced by 41%.
              • I’m confused. This is within the 95% confidence interval, but the accompanying chart has a p-value of 0.19. I’ll need to look up this result.
  • AHA Meta-Analysis of these four studies: AHA “performed a fixed-effects met-analysis of these 4 core trials.”
    • Result: Lowering saturated fat and replacing it with PUFA-rich vegetable oil lowed coronary heart disease by 29%.
      • I have to say, I do not understand their analysis at all. Especially from looking at the chart they provided. I’m not familiar with ‘inverse-variance fixed-effects meta-analysis,’ and I do not know how to interpret the chart. And do not see how the chart translates to their finding (29% reduction in CHD).


Initial thoughts

I have a lot of questions to answer, for which I’ll need to go look at the original studies. And do some additional background research probably.

Some of the criticisms I’ve seen to this include:

  • AHA ‘cherry-picked’ the studies. Well. AHA (presumably) identified criteria and then identified which studies met those criteria. To determine whether I think this is cherry-picking, I need to better understand the criteria and their value, and I also need to review the other studies and confirm that I agree they didn’t meet the criteria. For me to agree that AHA cherry-picked these studies, I would need some evidence or rationale that indicated that AHA chose the studies first and then chose criteria that matched those studies, rather than the reverse. We’ll see.
  • The studies are old. Well. So what? Good science is good science (and bad science is bad science). If something was done really well, why re-do it? This doesn’t necessarily mean the studies aren’t valuable – unless something else has changed since then that would invalidate these findings. I’m not sure if there is.

Generally, it seems like there are some meaningful, valid results. But I especially want to understand the implication of the durations more. These sound like long study durations… but if the same individuals weren’t in them for the full length, is that still as valid? And I’m also not convinced that the participants actually adhered to their assigned diets! Maybe I should be – maybe reduction in serum cholesterol is evidence of this. I just don’t know yet.

This might be the last you hear me talk about this, or it might not. Next steps are to continue summarizing the article and start to answer my own questions. Cheers.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s